High-Performance & eGPU

Ryzen 7 8845HS / 7840HS Beast

Zen 4 Hawk Point with Radeon 780M, a Ryzen AI NPU and — on many models — OCuLink. Native 1080p gaming today, desktop-GPU performance via eGPU tomorrow.

Recommended for Gaming Local AI & LLMs

ProcessorAMD Zen 4 Hawk Point (8C/16T, 5.1GHz)

GraphicsRadeon 780M (12 CUs RDNA 3)

Idle Power11W – 16W

Est. Barebone$490 - $680

Mini PCs with this chip

GMKtec

GMKtec K8 Plus

AMD Ryzen 7 8845HS (8C/16T, up to 5.1 GHz)

Graphics	AMD Radeon 780M (12 CUs, RDNA 3, up to 2.7 GHz)
Memory	2x DDR5-5600 SODIMM (Up to 128GB)
Storage	2x M.2 2280 PCIe 4.0 (Up to 16TB)
Idle Power	13W – 15W
Price	$409.99 barebone · $919.99 (32GB/1TB)

Chip Category

Ryzen 7 8845HS / 7840HS Beast

Full Specifications

Processor

Model	AMD Ryzen 7 8845HS (Hawk Point)
Cores / Threads	8 / 16 (Zen 4)
Base / Boost	3.8 GHz / 5.1 GHz
L3 Cache	16 MB
Default TDP	45 W (configurable 35–54 W)
NPU	AMD Ryzen AI (XDNA), up to 16 TOPS

Graphics & Media

iGPU	Radeon 780M, 12 CUs RDNA 3, up to 2.7 GHz
Media Engine	H.264 / HEVC / AV1 encode and decode
UMA Frame Buffer	BIOS-allocatable VRAM (up to 16 GB on many boards)

Memory & Storage

Memory Channels	Dual-channel
Memory Type	DDR5-5600 SODIMM or LPDDR5x-7500
Max Memory	96 GB (2x 48 GB SODIMM) on supporting boards
Storage	2x M.2 2280 PCIe 4.0 x4

Connectivity & Power

eGPU	OCuLink SFF-8611 (PCIe 4.0 x4) on selected models
USB4	40 Gbps, DisplayPort Alt-Mode + PD
Ethernet	1–2x 2.5GbE
Typical Idle	11 – 16 W at the wall
Power Input	19 V DC brick (120 W typical)

Strengths

Radeon 780M is one of the fastest integrated GPUs on the market
Hardware AV1 encode for streaming and transcoding
OCuLink gives near-native PCIe 4.0 x4 for an external desktop GPU
Shared memory makes 7B–14B quantised LLMs usable without a dGPU

Trade-offs

Higher idle draw than Intel N-series units
Some units use liquid metal TIM — avoid shaking or transporting vertically
OCuLink is not hot-pluggable; connect the dock with the PC powered off

Best Suited For

Native 1080p AAA gaming, external eGPU docking, Ollama local 7B-13B LLM inference.

Check before you buy

Verify whether OCuLink is a real port or an internal M.2-to-OCuLink adapter.
Buy matched dual-channel DDR5-5600 (2x 16 GB or 2x 32 GB).
Check the BIOS lets you raise the UMA frame buffer if you plan to run LLMs.

Tutorials for this hardware

Run Local LLMs with Ollama on the Radeon 780M · 30 minutes

Compare with

Flagship AI & Workstation

Strix Point Ryzen AI 9 HX 370

AMD Zen 5 (12C/24T, 50 TOPS NPU)

Graphics (iGPU)	Radeon 890M (16 CUs)
Memory Architecture	LPDDR5X 7500MT/s (Soldered 32/64GB)
Storage Exp.	Dual M.2 PCIe 4.0 x4
Idle Power	12W – 19W
Est. Barebone Cost	$790 - $1,050

Best Suited For

Ultra-fast local model fine-tuning, 1440p gaming, intensive creative studio workflow.

Flagship AI & Workstation

Gorgon Point Ryzen AI 9 HX 470

AMD Zen 5 (12C/24T, 55 TOPS NPU)

Graphics (iGPU)	Radeon 890M (16 CUs, up to 3.1 GHz)
Memory Architecture	DDR5-5600 SODIMM or LPDDR5X (Up to 128GB)
Storage Exp.	2–3x M.2 PCIe 4.0 x4 (chassis dependent)
Idle Power	6W – 10W
Est. Barebone Cost	$720 - $1,400

Best Suited For

Local LLM inference, 1080p–1440p gaming with upscaling, software builds and creative workloads in a compact desktop.

Unified-Memory AI Workstation

Strix Halo Ryzen AI Max+ 395

AMD Zen 5 (16C/32T, 50 TOPS NPU)

Graphics (iGPU)	Radeon 8060S (40 CUs, up to 2.9 GHz)
Memory Architecture	LPDDR5X-8000 256-bit, soldered (64GB or 128GB)
Storage Exp.	2x M.2 PCIe 4.0 x4
Idle Power	10W – 18W
Est. Barebone Cost	$1,500 - $2,800

Best Suited For

Local LLM inference on 30B–120B models, ROCm/AI development, 1440p gaming without a discrete GPU, heavy multi-core builds.